The Spring Framework provides extensive support for integrating with messaging systems, from simplified use of the JMS API using JmsTemplate to a complete infrastructure to receive messages asynchronously. Spring AMQP provides a similar feature set for the Advanced Message Queuing Protocol. Spring Boot also provides auto-configuration options for RabbitTemplate and RabbitMQ. Spring WebSocket natively includes support for STOMP messaging, and Spring Boot has support for that through starters and a small amount of auto-configuration. Spring Boot also has support for Apache Kafka.

1. JMS

The javax.jms.ConnectionFactory interface provides a standard method of creating a javax.jms.Connection for interacting with a JMS broker. Although Spring needs a ConnectionFactory to work with JMS, you generally need not use it directly yourself and can instead rely on higher level messaging abstractions. (See the relevant section of the Spring Framework reference documentation for details.) Spring Boot also auto-configures the necessary infrastructure to send and receive messages.

1.1. ActiveMQ Support

When ActiveMQ is available on the classpath, Spring Boot can also configure a ConnectionFactory. If the broker is present, an embedded broker is automatically started and configured (provided no broker URL is specified through configuration and the embedded broker is not disabled in the configuration).

If you use spring-boot-starter-activemq, the necessary dependencies to connect or embed an ActiveMQ instance are provided, as is the Spring infrastructure to integrate with JMS.

ActiveMQ configuration is controlled by external configuration properties in spring.activemq.*.

By default, ActiveMQ is auto-configured to use the VM transport, which starts a broker embedded in the same JVM instance.

You can disable the embedded broker by configuring the spring.activemq.in-memory property, as shown in the following example:

Properties
spring.activemq.in-memory=false
Yaml
spring:
  activemq:
    in-memory: false

The embedded broker will also be disabled if you configure the broker URL, as shown in the following example:

Properties
spring.activemq.broker-url=tcp://192.168.1.210:9876
spring.activemq.user=admin
spring.activemq.password=secret
Yaml
spring:
  activemq:
    broker-url: "tcp://192.168.1.210:9876"
    user: "admin"
    password: "secret"

If you want to take full control over the embedded broker, see the ActiveMQ documentation for further information.

By default, a CachingConnectionFactory wraps the native ConnectionFactory with sensible settings that you can control by external configuration properties in spring.jms.*:

Properties
spring.jms.cache.session-cache-size=5
Yaml
spring:
  jms:
    cache:
      session-cache-size: 5

If you’d rather use native pooling, you can do so by adding a dependency to org.messaginghub:pooled-jms and configuring the JmsPoolConnectionFactory accordingly, as shown in the following example:

Properties
spring.activemq.pool.enabled=true
spring.activemq.pool.max-connections=50
Yaml
spring:
  activemq:
    pool:
      enabled: true
      max-connections: 50
See ActiveMQProperties for more of the supported options. You can also register an arbitrary number of beans that implement ActiveMQConnectionFactoryCustomizer for more advanced customizations.

By default, ActiveMQ creates a destination if it does not yet exist so that destinations are resolved against their provided names.

1.2. ActiveMQ Artemis Support

Spring Boot can auto-configure a ConnectionFactory when it detects that ActiveMQ Artemis is available on the classpath. If the broker is present, an embedded broker is automatically started and configured (unless the mode property has been explicitly set). The supported modes are embedded (to make explicit that an embedded broker is required and that an error should occur if the broker is not available on the classpath) and native (to connect to a broker using the netty transport protocol). When the latter is configured, Spring Boot configures a ConnectionFactory that connects to a broker running on the local machine with the default settings.

If you use spring-boot-starter-artemis, the necessary dependencies to connect to an existing ActiveMQ Artemis instance are provided, as well as the Spring infrastructure to integrate with JMS. Adding org.apache.activemq:artemis-jms-server to your application lets you use embedded mode.

ActiveMQ Artemis configuration is controlled by external configuration properties in spring.artemis.*. For example, you might declare the following section in application.properties:

Properties
spring.artemis.mode=native
spring.artemis.broker-url=tcp://192.168.1.210:9876
spring.artemis.user=admin
spring.artemis.password=secret
Yaml
spring:
  artemis:
    mode: native
    broker-url: "tcp://192.168.1.210:9876"
    user: "admin"
    password: "secret"

When embedding the broker, you can choose if you want to enable persistence and list the destinations that should be made available. These can be specified as a comma-separated list to create them with the default options, or you can define bean(s) of type org.apache.activemq.artemis.jms.server.config.JMSQueueConfiguration or org.apache.activemq.artemis.jms.server.config.TopicConfiguration, for advanced queue and topic configurations, respectively.

By default, a CachingConnectionFactory wraps the native ConnectionFactory with sensible settings that you can control by external configuration properties in spring.jms.*:

Properties
spring.jms.cache.session-cache-size=5
Yaml
spring:
  jms:
    cache:
      session-cache-size: 5

If you’d rather use native pooling, you can do so by adding a dependency to org.messaginghub:pooled-jms and configuring the JmsPoolConnectionFactory accordingly, as shown in the following example:

Properties
spring.artemis.pool.enabled=true
spring.artemis.pool.max-connections=50
Yaml
spring:
  artemis:
    pool:
      enabled: true
      max-connections: 50

See ArtemisProperties for more supported options.

No JNDI lookup is involved, and destinations are resolved against their names, using either the name attribute in the Artemis configuration or the names provided through configuration.

1.3. Using a JNDI ConnectionFactory

If you are running your application in an application server, Spring Boot tries to locate a JMS ConnectionFactory by using JNDI. By default, the java:/JmsXA and java:/XAConnectionFactory location are checked. You can use the spring.jms.jndi-name property if you need to specify an alternative location, as shown in the following example:

Properties
spring.jms.jndi-name=java:/MyConnectionFactory
Yaml
spring:
  jms:
    jndi-name: "java:/MyConnectionFactory"

1.4. Sending a Message

Spring’s JmsTemplate is auto-configured, and you can autowire it directly into your own beans, as shown in the following example:

Java
import org.springframework.jms.core.JmsTemplate;
import org.springframework.stereotype.Component;

@Component
public class MyBean {

    private final JmsTemplate jmsTemplate;

    public MyBean(JmsTemplate jmsTemplate) {
        this.jmsTemplate = jmsTemplate;
    }

    // ...

    public void someMethod() {
        this.jmsTemplate.convertAndSend("hello");
    }

}
Kotlin
import org.springframework.jms.core.JmsTemplate
import org.springframework.stereotype.Component

@Component
class MyBean(private val jmsTemplate: JmsTemplate) {

    // ...

    fun someMethod() {
        jmsTemplate.convertAndSend("hello")
    }

}
JmsMessagingTemplate can be injected in a similar manner. If a DestinationResolver or a MessageConverter bean is defined, it is associated automatically to the auto-configured JmsTemplate.

1.5. Receiving a Message

When the JMS infrastructure is present, any bean can be annotated with @JmsListener to create a listener endpoint. If no JmsListenerContainerFactory has been defined, a default one is configured automatically. If a DestinationResolver, a MessageConverter, or a javax.jms.ExceptionListener beans are defined, they are associated automatically with the default factory.

By default, the default factory is transactional. If you run in an infrastructure where a JtaTransactionManager is present, it is associated to the listener container by default. If not, the sessionTransacted flag is enabled. In that latter scenario, you can associate your local data store transaction to the processing of an incoming message by adding @Transactional on your listener method (or a delegate thereof). This ensures that the incoming message is acknowledged, once the local transaction has completed. This also includes sending response messages that have been performed on the same JMS session.

The following component creates a listener endpoint on the someQueue destination:

Java
import org.springframework.jms.annotation.JmsListener;
import org.springframework.stereotype.Component;

@Component
public class MyBean {

    @JmsListener(destination = "someQueue")
    public void processMessage(String content) {
        // ...
    }

}
Kotlin
import org.springframework.jms.annotation.JmsListener
import org.springframework.stereotype.Component

@Component
class MyBean {

    @JmsListener(destination = "someQueue")
    fun processMessage(content: String?) {
        // ...
    }

}
See the Javadoc of @EnableJms for more details.

If you need to create more JmsListenerContainerFactory instances or if you want to override the default, Spring Boot provides a DefaultJmsListenerContainerFactoryConfigurer that you can use to initialize a DefaultJmsListenerContainerFactory with the same settings as the one that is auto-configured.

For instance, the following example exposes another factory that uses a specific MessageConverter:

Java
import javax.jms.ConnectionFactory;

import org.springframework.boot.autoconfigure.jms.DefaultJmsListenerContainerFactoryConfigurer;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.jms.config.DefaultJmsListenerContainerFactory;

@Configuration(proxyBeanMethods = false)
public class MyJmsConfiguration {

    @Bean
    public DefaultJmsListenerContainerFactory myFactory(DefaultJmsListenerContainerFactoryConfigurer configurer) {
        DefaultJmsListenerContainerFactory factory = new DefaultJmsListenerContainerFactory();
        ConnectionFactory connectionFactory = getCustomConnectionFactory();
        configurer.configure(factory, connectionFactory);
        factory.setMessageConverter(new MyMessageConverter());
        return factory;
    }

    private ConnectionFactory getCustomConnectionFactory() {
        return ...
    }

}
Kotlin
import org.springframework.boot.autoconfigure.jms.DefaultJmsListenerContainerFactoryConfigurer
import org.springframework.context.annotation.Bean
import org.springframework.context.annotation.Configuration
import org.springframework.jms.config.DefaultJmsListenerContainerFactory
import javax.jms.ConnectionFactory

@Configuration(proxyBeanMethods = false)
class MyJmsConfiguration {

    @Bean
    fun myFactory(configurer: DefaultJmsListenerContainerFactoryConfigurer): DefaultJmsListenerContainerFactory {
        val factory = DefaultJmsListenerContainerFactory()
        val connectionFactory = getCustomConnectionFactory()
        configurer.configure(factory, connectionFactory)
        factory.setMessageConverter(MyMessageConverter())
        return factory
    }

    fun getCustomConnectionFactory() : ConnectionFactory? {
        return ...
    }

}

Then you can use the factory in any @JmsListener-annotated method as follows:

Java
import org.springframework.jms.annotation.JmsListener;
import org.springframework.stereotype.Component;

@Component
public class MyBean {

    @JmsListener(destination = "someQueue", containerFactory = "myFactory")
    public void processMessage(String content) {
        // ...
    }

}
Kotlin
import org.springframework.jms.annotation.JmsListener
import org.springframework.stereotype.Component

@Component
class MyBean {

    @JmsListener(destination = "someQueue", containerFactory = "myFactory")
    fun processMessage(content: String?) {
        // ...
    }

}

2. AMQP

The Advanced Message Queuing Protocol (AMQP) is a platform-neutral, wire-level protocol for message-oriented middleware. The Spring AMQP project applies core Spring concepts to the development of AMQP-based messaging solutions. Spring Boot offers several conveniences for working with AMQP through RabbitMQ, including the spring-boot-starter-amqp “Starter”.

2.1. RabbitMQ Support

RabbitMQ is a lightweight, reliable, scalable, and portable message broker based on the AMQP protocol. Spring uses RabbitMQ to communicate through the AMQP protocol.

RabbitMQ configuration is controlled by external configuration properties in spring.rabbitmq.*. For example, you might declare the following section in application.properties:

Properties
spring.rabbitmq.host=localhost
spring.rabbitmq.port=5672
spring.rabbitmq.username=admin
spring.rabbitmq.password=secret
Yaml
spring:
  rabbitmq:
    host: "localhost"
    port: 5672
    username: "admin"
    password: "secret"

Alternatively, you could configure the same connection using the addresses attribute:

Properties
spring.rabbitmq.addresses=amqp://admin:secret@localhost
Yaml
spring:
  rabbitmq:
    addresses: "amqp://admin:secret@localhost"
When specifying addresses that way, the host and port properties are ignored. If the address uses the amqps protocol, SSL support is enabled automatically.

See RabbitProperties for more of the supported property-based configuration options. To configure lower-level details of the RabbitMQ ConnectionFactory that is used by Spring AMQP, define a ConnectionFactoryCustomizer bean.

If a ConnectionNameStrategy bean exists in the context, it will be automatically used to name connections created by the auto-configured CachingConnectionFactory.

2.2. Sending a Message

Spring’s AmqpTemplate and AmqpAdmin are auto-configured, and you can autowire them directly into your own beans, as shown in the following example:

Java
import org.springframework.amqp.core.AmqpAdmin;
import org.springframework.amqp.core.AmqpTemplate;
import org.springframework.stereotype.Component;

@Component
public class MyBean {

    private final AmqpAdmin amqpAdmin;

    private final AmqpTemplate amqpTemplate;

    public MyBean(AmqpAdmin amqpAdmin, AmqpTemplate amqpTemplate) {
        this.amqpAdmin = amqpAdmin;
        this.amqpTemplate = amqpTemplate;
    }

    // ...

    public void someMethod() {
        this.amqpAdmin.getQueueInfo("someQueue");
    }

    public void someOtherMethod() {
        this.amqpTemplate.convertAndSend("hello");
    }

}
Kotlin
import org.springframework.amqp.core.AmqpAdmin
import org.springframework.amqp.core.AmqpTemplate
import org.springframework.stereotype.Component

@Component
class MyBean(private val amqpAdmin: AmqpAdmin, private val amqpTemplate: AmqpTemplate) {

    // ...

    fun someMethod() {
        amqpAdmin.getQueueInfo("someQueue")
    }

    fun someOtherMethod() {
        amqpTemplate.convertAndSend("hello")
    }

}
RabbitMessagingTemplate can be injected in a similar manner. If a MessageConverter bean is defined, it is associated automatically to the auto-configured AmqpTemplate.

If necessary, any org.springframework.amqp.core.Queue that is defined as a bean is automatically used to declare a corresponding queue on the RabbitMQ instance.

To retry operations, you can enable retries on the AmqpTemplate (for example, in the event that the broker connection is lost):

Properties
spring.rabbitmq.template.retry.enabled=true
spring.rabbitmq.template.retry.initial-interval=2s
Yaml
spring:
  rabbitmq:
    template:
      retry:
        enabled: true
        initial-interval: "2s"

Retries are disabled by default. You can also customize the RetryTemplate programmatically by declaring a RabbitRetryTemplateCustomizer bean.

If you need to create more RabbitTemplate instances or if you want to override the default, Spring Boot provides a RabbitTemplateConfigurer bean that you can use to initialize a RabbitTemplate with the same settings as the factories used by the auto-configuration.

2.3. Sending a Message To A Stream

To send a message to a particular stream, specify the name of the stream, as shown in the following example:

Properties
spring.rabbitmq.stream.name=my-stream
Yaml
spring:
  rabbitmq:
    stream:
      name: "my-stream"

If a MessageConverter, StreamMessageConverter, or ProducerCustomizer bean is defined, it is associated automatically to the auto-configured RabbitStreamTemplate.

If you need to create more RabbitStreamTemplate instances or if you want to override the default, Spring Boot provides a RabbitStreamTemplateConfigurer bean that you can use to initialize a RabbitStreamTemplate with the same settings as the factories used by the auto-configuration.

2.4. Receiving a Message

When the Rabbit infrastructure is present, any bean can be annotated with @RabbitListener to create a listener endpoint. If no RabbitListenerContainerFactory has been defined, a default SimpleRabbitListenerContainerFactory is automatically configured and you can switch to a direct container using the spring.rabbitmq.listener.type property. If a MessageConverter or a MessageRecoverer bean is defined, it is automatically associated with the default factory.

The following sample component creates a listener endpoint on the someQueue queue:

Java
import org.springframework.amqp.rabbit.annotation.RabbitListener;
import org.springframework.stereotype.Component;

@Component
public class MyBean {

    @RabbitListener(queues = "someQueue")
    public void processMessage(String content) {
        // ...
    }

}
Kotlin
import org.springframework.amqp.rabbit.annotation.RabbitListener
import org.springframework.stereotype.Component

@Component
class MyBean {

    @RabbitListener(queues = ["someQueue"])
    fun processMessage(content: String?) {
        // ...
    }

}
See the Javadoc of @EnableRabbit for more details.

If you need to create more RabbitListenerContainerFactory instances or if you want to override the default, Spring Boot provides a SimpleRabbitListenerContainerFactoryConfigurer and a DirectRabbitListenerContainerFactoryConfigurer that you can use to initialize a SimpleRabbitListenerContainerFactory and a DirectRabbitListenerContainerFactory with the same settings as the factories used by the auto-configuration.

It does not matter which container type you chose. Those two beans are exposed by the auto-configuration.

For instance, the following configuration class exposes another factory that uses a specific MessageConverter:

Java
import org.springframework.amqp.rabbit.config.SimpleRabbitListenerContainerFactory;
import org.springframework.amqp.rabbit.connection.ConnectionFactory;
import org.springframework.boot.autoconfigure.amqp.SimpleRabbitListenerContainerFactoryConfigurer;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;

@Configuration(proxyBeanMethods = false)
public class MyRabbitConfiguration {

    @Bean
    public SimpleRabbitListenerContainerFactory myFactory(SimpleRabbitListenerContainerFactoryConfigurer configurer) {
        SimpleRabbitListenerContainerFactory factory = new SimpleRabbitListenerContainerFactory();
        ConnectionFactory connectionFactory = getCustomConnectionFactory();
        configurer.configure(factory, connectionFactory);
        factory.setMessageConverter(new MyMessageConverter());
        return factory;
    }

    private ConnectionFactory getCustomConnectionFactory() {
        return ...
    }

}
Kotlin
import org.springframework.amqp.rabbit.config.SimpleRabbitListenerContainerFactory
import org.springframework.amqp.rabbit.connection.ConnectionFactory
import org.springframework.boot.autoconfigure.amqp.SimpleRabbitListenerContainerFactoryConfigurer
import org.springframework.context.annotation.Bean
import org.springframework.context.annotation.Configuration

@Configuration(proxyBeanMethods = false)
class MyRabbitConfiguration {

    @Bean
    fun myFactory(configurer: SimpleRabbitListenerContainerFactoryConfigurer): SimpleRabbitListenerContainerFactory {
        val factory = SimpleRabbitListenerContainerFactory()
        val connectionFactory = getCustomConnectionFactory()
        configurer.configure(factory, connectionFactory)
        factory.setMessageConverter(MyMessageConverter())
        return factory
    }

    fun getCustomConnectionFactory() : ConnectionFactory? {
        return ...
    }

}

Then you can use the factory in any @RabbitListener-annotated method, as follows:

Java
import org.springframework.amqp.rabbit.annotation.RabbitListener;
import org.springframework.stereotype.Component;

@Component
public class MyBean {

    @RabbitListener(queues = "someQueue", containerFactory = "myFactory")
    public void processMessage(String content) {
        // ...
    }

}
Kotlin
import org.springframework.amqp.rabbit.annotation.RabbitListener
import org.springframework.stereotype.Component

@Component
class MyBean {

    @RabbitListener(queues = ["someQueue"], containerFactory = "myFactory")
    fun processMessage(content: String?) {
        // ...
    }

}

You can enable retries to handle situations where your listener throws an exception. By default, RejectAndDontRequeueRecoverer is used, but you can define a MessageRecoverer of your own. When retries are exhausted, the message is rejected and either dropped or routed to a dead-letter exchange if the broker is configured to do so. By default, retries are disabled. You can also customize the RetryTemplate programmatically by declaring a RabbitRetryTemplateCustomizer bean.

By default, if retries are disabled and the listener throws an exception, the delivery is retried indefinitely. You can modify this behavior in two ways: Set the defaultRequeueRejected property to false so that zero re-deliveries are attempted or throw an AmqpRejectAndDontRequeueException to signal the message should be rejected. The latter is the mechanism used when retries are enabled and the maximum number of delivery attempts is reached.

3. Apache Kafka Support

Apache Kafka is supported by providing auto-configuration of the spring-kafka project.

Kafka configuration is controlled by external configuration properties in spring.kafka.*. For example, you might declare the following section in application.properties:

Properties
spring.kafka.bootstrap-servers=localhost:9092
spring.kafka.consumer.group-id=myGroup
Yaml
spring:
  kafka:
    bootstrap-servers: "localhost:9092"
    consumer:
      group-id: "myGroup"
To create a topic on startup, add a bean of type NewTopic. If the topic already exists, the bean is ignored.

See KafkaProperties for more supported options.

3.1. Sending a Message

Spring’s KafkaTemplate is auto-configured, and you can autowire it directly in your own beans, as shown in the following example:

Java
import org.springframework.kafka.core.KafkaTemplate;
import org.springframework.stereotype.Component;

@Component
public class MyBean {

    private final KafkaTemplate<String, String> kafkaTemplate;

    public MyBean(KafkaTemplate<String, String> kafkaTemplate) {
        this.kafkaTemplate = kafkaTemplate;
    }

    // ...

    public void someMethod() {
        this.kafkaTemplate.send("someTopic", "Hello");
    }

}
Kotlin
import org.springframework.kafka.core.KafkaTemplate
import org.springframework.stereotype.Component

@Component
class MyBean(private val kafkaTemplate: KafkaTemplate<String, String>) {

    // ...

    fun someMethod() {
        kafkaTemplate.send("someTopic", "Hello")
    }

}
If the property spring.kafka.producer.transaction-id-prefix is defined, a KafkaTransactionManager is automatically configured. Also, if a RecordMessageConverter bean is defined, it is automatically associated to the auto-configured KafkaTemplate.

3.2. Receiving a Message

When the Apache Kafka infrastructure is present, any bean can be annotated with @KafkaListener to create a listener endpoint. If no KafkaListenerContainerFactory has been defined, a default one is automatically configured with keys defined in spring.kafka.listener.*.

The following component creates a listener endpoint on the someTopic topic:

Java
import org.springframework.kafka.annotation.KafkaListener;
import org.springframework.stereotype.Component;

@Component
public class MyBean {

    @KafkaListener(topics = "someTopic")
    public void processMessage(String content) {
        // ...
    }

}
Kotlin
import org.springframework.kafka.annotation.KafkaListener
import org.springframework.stereotype.Component

@Component
class MyBean {

    @KafkaListener(topics = ["someTopic"])
    fun processMessage(content: String?) {
        // ...
    }

}

If a KafkaTransactionManager bean is defined, it is automatically associated to the container factory. Similarly, if a RecordFilterStrategy, CommonErrorHandler, AfterRollbackProcessor or ConsumerAwareRebalanceListener bean is defined, it is automatically associated to the default factory.

Depending on the listener type, a RecordMessageConverter or BatchMessageConverter bean is associated to the default factory. If only a RecordMessageConverter bean is present for a batch listener, it is wrapped in a BatchMessageConverter.

A custom ChainedKafkaTransactionManager must be marked @Primary as it usually references the auto-configured KafkaTransactionManager bean.

3.3. Kafka Streams

Spring for Apache Kafka provides a factory bean to create a StreamsBuilder object and manage the lifecycle of its streams. Spring Boot auto-configures the required KafkaStreamsConfiguration bean as long as kafka-streams is on the classpath and Kafka Streams is enabled by the @EnableKafkaStreams annotation.

Enabling Kafka Streams means that the application id and bootstrap servers must be set. The former can be configured using spring.kafka.streams.application-id, defaulting to spring.application.name if not set. The latter can be set globally or specifically overridden only for streams.

Several additional properties are available using dedicated properties; other arbitrary Kafka properties can be set using the spring.kafka.streams.properties namespace. See also Additional Kafka Properties for more information.

To use the factory bean, wire StreamsBuilder into your @Bean as shown in the following example:

Java
import org.apache.kafka.common.serialization.Serdes;
import org.apache.kafka.streams.KeyValue;
import org.apache.kafka.streams.StreamsBuilder;
import org.apache.kafka.streams.kstream.KStream;
import org.apache.kafka.streams.kstream.Produced;

import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.kafka.annotation.EnableKafkaStreams;
import org.springframework.kafka.support.serializer.JsonSerde;

@Configuration(proxyBeanMethods = false)
@EnableKafkaStreams
public class MyKafkaStreamsConfiguration {

    @Bean
    public KStream<Integer, String> kStream(StreamsBuilder streamsBuilder) {
        KStream<Integer, String> stream = streamsBuilder.stream("ks1In");
        stream.map(this::uppercaseValue).to("ks1Out", Produced.with(Serdes.Integer(), new JsonSerde<>()));
        return stream;
    }

    private KeyValue<Integer, String> uppercaseValue(Integer key, String value) {
        return new KeyValue<>(key, value.toUpperCase());
    }

}
Kotlin
import org.apache.kafka.common.serialization.Serdes
import org.apache.kafka.streams.KeyValue
import org.apache.kafka.streams.StreamsBuilder
import org.apache.kafka.streams.kstream.KStream
import org.apache.kafka.streams.kstream.Produced
import org.springframework.context.annotation.Bean
import org.springframework.context.annotation.Configuration
import org.springframework.kafka.annotation.EnableKafkaStreams
import org.springframework.kafka.support.serializer.JsonSerde

@Configuration(proxyBeanMethods = false)
@EnableKafkaStreams
class MyKafkaStreamsConfiguration {

    @Bean
    fun kStream(streamsBuilder: StreamsBuilder): KStream<Int, String> {
        val stream = streamsBuilder.stream<Int, String>("ks1In")
        stream.map(this::uppercaseValue).to("ks1Out", Produced.with(Serdes.Integer(), JsonSerde()))
        return stream
    }

    private fun uppercaseValue(key: Int, value: String): KeyValue<Int?, String?> {
        return KeyValue(key, value.uppercase())
    }

}

By default, the streams managed by the StreamBuilder object are started automatically. You can customize this behavior using the spring.kafka.streams.auto-startup property.

3.4. Additional Kafka Properties

The properties supported by auto configuration are shown in the “Integration Properties” section of the Appendix. Note that, for the most part, these properties (hyphenated or camelCase) map directly to the Apache Kafka dotted properties. See the Apache Kafka documentation for details.

The first few of these properties apply to all components (producers, consumers, admins, and streams) but can be specified at the component level if you wish to use different values. Apache Kafka designates properties with an importance of HIGH, MEDIUM, or LOW. Spring Boot auto-configuration supports all HIGH importance properties, some selected MEDIUM and LOW properties, and any properties that do not have a default value.

Only a subset of the properties supported by Kafka are available directly through the KafkaProperties class. If you wish to configure the producer or consumer with additional properties that are not directly supported, use the following properties:

Properties
spring.kafka.properties[prop.one]=first
spring.kafka.admin.properties[prop.two]=second
spring.kafka.consumer.properties[prop.three]=third
spring.kafka.producer.properties[prop.four]=fourth
spring.kafka.streams.properties[prop.five]=fifth
Yaml
spring:
  kafka:
    properties:
      "[prop.one]": "first"
    admin:
      properties:
        "[prop.two]": "second"
    consumer:
      properties:
        "[prop.three]": "third"
    producer:
      properties:
        "[prop.four]": "fourth"
    streams:
      properties:
        "[prop.five]": "fifth"

This sets the common prop.one Kafka property to first (applies to producers, consumers and admins), the prop.two admin property to second, the prop.three consumer property to third, the prop.four producer property to fourth and the prop.five streams property to fifth.

You can also configure the Spring Kafka JsonDeserializer as follows:

Properties
spring.kafka.consumer.value-deserializer=org.springframework.kafka.support.serializer.JsonDeserializer
spring.kafka.consumer.properties[spring.json.value.default.type]=com.example.Invoice
spring.kafka.consumer.properties[spring.json.trusted.packages]=com.example.main,com.example.another
Yaml
spring:
  kafka:
    consumer:
      value-deserializer: "org.springframework.kafka.support.serializer.JsonDeserializer"
      properties:
        "[spring.json.value.default.type]": "com.example.Invoice"
        "[spring.json.trusted.packages]": "com.example.main,com.example.another"

Similarly, you can disable the JsonSerializer default behavior of sending type information in headers:

Properties
spring.kafka.producer.value-serializer=org.springframework.kafka.support.serializer.JsonSerializer
spring.kafka.producer.properties[spring.json.add.type.headers]=false
Yaml
spring:
  kafka:
    producer:
      value-serializer: "org.springframework.kafka.support.serializer.JsonSerializer"
      properties:
        "[spring.json.add.type.headers]": false
Properties set in this way override any configuration item that Spring Boot explicitly supports.

3.5. Testing with Embedded Kafka

Spring for Apache Kafka provides a convenient way to test projects with an embedded Apache Kafka broker. To use this feature, annotate a test class with @EmbeddedKafka from the spring-kafka-test module. For more information, please see the Spring for Apache Kafka reference manual.

To make Spring Boot auto-configuration work with the aforementioned embedded Apache Kafka broker, you need to remap a system property for embedded broker addresses (populated by the EmbeddedKafkaBroker) into the Spring Boot configuration property for Apache Kafka. There are several ways to do that:

  • Provide a system property to map embedded broker addresses into spring.kafka.bootstrap-servers in the test class:

Java
static {
    System.setProperty(EmbeddedKafkaBroker.BROKER_LIST_PROPERTY, "spring.kafka.bootstrap-servers");
}
Kotlin
init {
    System.setProperty(EmbeddedKafkaBroker.BROKER_LIST_PROPERTY, "spring.kafka.bootstrap-servers")
}
  • Configure a property name on the @EmbeddedKafka annotation:

Java
import org.springframework.boot.test.context.SpringBootTest;
import org.springframework.kafka.test.context.EmbeddedKafka;

@SpringBootTest
@EmbeddedKafka(topics = "someTopic", bootstrapServersProperty = "spring.kafka.bootstrap-servers")
class MyTest {

    // ...

}
Kotlin
import org.springframework.boot.test.context.SpringBootTest
import org.springframework.kafka.test.context.EmbeddedKafka

@SpringBootTest
@EmbeddedKafka(topics = ["someTopic"], bootstrapServersProperty = "spring.kafka.bootstrap-servers")
class MyTest {

    // ...

}
  • Use a placeholder in configuration properties:

Properties
spring.kafka.bootstrap-servers=${spring.embedded.kafka.brokers}
Yaml
spring:
  kafka:
    bootstrap-servers: "${spring.embedded.kafka.brokers}"

4. RSocket

RSocket is a binary protocol for use on byte stream transports. It enables symmetric interaction models through async message passing over a single connection.

The spring-messaging module of the Spring Framework provides support for RSocket requesters and responders, both on the client and on the server side. See the RSocket section of the Spring Framework reference for more details, including an overview of the RSocket protocol.

4.1. RSocket Strategies Auto-configuration

Spring Boot auto-configures an RSocketStrategies bean that provides all the required infrastructure for encoding and decoding RSocket payloads. By default, the auto-configuration will try to configure the following (in order):

  1. CBOR codecs with Jackson

  2. JSON codecs with Jackson

The spring-boot-starter-rsocket starter provides both dependencies. See the Jackson support section to know more about customization possibilities.

Developers can customize the RSocketStrategies component by creating beans that implement the RSocketStrategiesCustomizer interface. Note that their @Order is important, as it determines the order of codecs.

4.2. RSocket server Auto-configuration

Spring Boot provides RSocket server auto-configuration. The required dependencies are provided by the spring-boot-starter-rsocket.

Spring Boot allows exposing RSocket over WebSocket from a WebFlux server, or standing up an independent RSocket server. This depends on the type of application and its configuration.

For WebFlux application (that is of type WebApplicationType.REACTIVE), the RSocket server will be plugged into the Web Server only if the following properties match:

Properties
spring.rsocket.server.mapping-path=/rsocket
spring.rsocket.server.transport=websocket
Yaml
spring:
  rsocket:
    server:
      mapping-path: "/rsocket"
      transport: "websocket"
Plugging RSocket into a web server is only supported with Reactor Netty, as RSocket itself is built with that library.

Alternatively, an RSocket TCP or websocket server is started as an independent, embedded server. Besides the dependency requirements, the only required configuration is to define a port for that server:

Properties
spring.rsocket.server.port=9898
Yaml
spring:
  rsocket:
    server:
      port: 9898

4.3. Spring Messaging RSocket support

Spring Boot will auto-configure the Spring Messaging infrastructure for RSocket.

This means that Spring Boot will create a RSocketMessageHandler bean that will handle RSocket requests to your application.

4.4. Calling RSocket Services with RSocketRequester

Once the RSocket channel is established between server and client, any party can send or receive requests to the other.

As a server, you can get injected with an RSocketRequester instance on any handler method of an RSocket @Controller. As a client, you need to configure and establish an RSocket connection first. Spring Boot auto-configures an RSocketRequester.Builder for such cases with the expected codecs and applies any RSocketConnectorConfigurer bean.

The RSocketRequester.Builder instance is a prototype bean, meaning each injection point will provide you with a new instance . This is done on purpose since this builder is stateful and you should not create requesters with different setups using the same instance.

The following code shows a typical example:

Java
import reactor.core.publisher.Mono;

import org.springframework.messaging.rsocket.RSocketRequester;
import org.springframework.stereotype.Service;

@Service
public class MyService {

    private final RSocketRequester rsocketRequester;

    public MyService(RSocketRequester.Builder rsocketRequesterBuilder) {
        this.rsocketRequester = rsocketRequesterBuilder.tcp("example.org", 9898);
    }

    public Mono<User> someRSocketCall(String name) {
        return this.rsocketRequester.route("user").data(name).retrieveMono(User.class);
    }

}
Kotlin
import org.springframework.messaging.rsocket.RSocketRequester
import org.springframework.stereotype.Service
import reactor.core.publisher.Mono

@Service
class MyService(rsocketRequesterBuilder: RSocketRequester.Builder) {

    private val rsocketRequester: RSocketRequester

    init {
        rsocketRequester = rsocketRequesterBuilder.tcp("example.org", 9898)
    }

    fun someRSocketCall(name: String): Mono<User> {
        return rsocketRequester.route("user").data(name).retrieveMono(
            User::class.java
        )
    }

}

5. Spring Integration

Spring Boot offers several conveniences for working with Spring Integration, including the spring-boot-starter-integration “Starter”. Spring Integration provides abstractions over messaging and also other transports such as HTTP, TCP, and others. If Spring Integration is available on your classpath, it is initialized through the @EnableIntegration annotation.

Spring Integration polling logic relies on the auto-configured TaskScheduler. The default PollerMetadata (poll unbounded number of messages every second) can be customized with spring.integration.poller.* configuration properties.

Spring Boot also configures some features that are triggered by the presence of additional Spring Integration modules. If spring-integration-jmx is also on the classpath, message processing statistics are published over JMX. If spring-integration-jdbc is available, the default database schema can be created on startup, as shown in the following line:

Properties
spring.integration.jdbc.initialize-schema=always
Yaml
spring:
  integration:
    jdbc:
      initialize-schema: "always"

If spring-integration-rsocket is available, developers can configure an RSocket server using "spring.rsocket.server.*" properties and let it use IntegrationRSocketEndpoint or RSocketOutboundGateway components to handle incoming RSocket messages. This infrastructure can handle Spring Integration RSocket channel adapters and @MessageMapping handlers (given "spring.integration.rsocket.server.message-mapping-enabled" is configured).

Spring Boot can also auto-configure an ClientRSocketConnector using configuration properties:

Properties
# Connecting to a RSocket server over TCP
spring.integration.rsocket.client.host=example.org
spring.integration.rsocket.client.port=9898
Yaml
# Connecting to a RSocket server over TCP
spring:
  integration:
    rsocket:
      client:
        host: "example.org"
        port: 9898
Properties
# Connecting to a RSocket Server over WebSocket
spring.integration.rsocket.client.uri=ws://example.org
Yaml
# Connecting to a RSocket Server over WebSocket
spring:
  integration:
    rsocket:
      client:
        uri: "ws://example.org"

See the IntegrationAutoConfiguration and IntegrationProperties classes for more details.

6. WebSockets

Spring Boot provides WebSockets auto-configuration for embedded Tomcat, Jetty, and Undertow. If you deploy a war file to a standalone container, Spring Boot assumes that the container is responsible for the configuration of its WebSocket support.

Spring Framework provides rich WebSocket support for MVC web applications that can be easily accessed through the spring-boot-starter-websocket module.

WebSocket support is also available for reactive web applications and requires to include the WebSocket API alongside spring-boot-starter-webflux:

<dependency>
    <groupId>javax.websocket</groupId>
    <artifactId>javax.websocket-api</artifactId>
</dependency>

7. What to Read Next

The next section describes how to enable IO capabilities in your application. You can read about caching, mail, validation, rest clients and more in this section.